Reality Is Built, with Inescapable Logic, from Bits

Quantum mechanics has a reputation, and it isn’t a good one. Particles that are in two places at once until you look. Cats that are alive and dead. Measurements that change what they measure. “Spooky action at a distance,” in Einstein’s exasperated phrase. The usual way the subject is taught does nothing to dispel the unease: here is a set of postulates — complex numbers, wavefunctions, a rule for turning them into probabilities, a mysterious “collapse” when you observe — and they are handed to you not because they make sense but because, against all intuition, they work.

This essay is about a different possibility, one that has been taking shape quietly in the foundations of physics for the last two decades. The possibility is this: quantum mechanics is not a weird set of postulates at all. It is what you are forced into, step by logical step, the moment you ask for a universe that can keep a record of anything. Not “happens to be.” Forced into. And if that’s right, then the strangeness was never in the physics. It was in our expectations — expectations formed in a corner of reality so unusual that it hid the rules from us.

I’m going to try to make that case carefully, defining every piece of jargon before I lean on it, and — this matters — being honest at each step about how solid the reasoning is. Some of the steps are airtight theorems. Some are strong arguments that rest on a reasonable assumption. A few, near the end, are frontier conjecture. I’ll mark them as we go, because a chain of reasoning is only as trustworthy as its weakest link is honest.

Let me start with the slogan, and then earn it.

Wheeler’s three words

In 1990 the physicist John Archibald Wheeler — Feynman’s mentor, the man who coined the term “black hole” — distilled a lifetime of puzzling over the quantum into three words: it from bit. His idea was that every physical thing (“it”) — every particle, every field, every patch of space — owes its existence, at bottom, to answers to yes-or-no questions. To information. Reality, he suspected, is made of bits.

It was a beautiful slogan and, for thirty years, mostly just a slogan. Inspiring, unprovable, the kind of thing you nod at without quite knowing what you’d do with it.

To do something with it, we need a tool that has matured only recently: reconstruction. The ordinary way to present a physical theory is to write down its rules and check them against experiment. Reconstruction runs the other way. It asks: from what simple, reasonable requirements do the rules follow, so that they stop looking arbitrary? This has been done, with real rigour, for quantum theory — by Lucien Hardy, who derived it from five “reasonable axioms”; by Giulio Chiribella and collaborators, who derived it from principles about information; by Wojciech Zurek, who showed that the ordinary classical world emerges naturally from quantum rules once you take copying seriously. The dream of the program is to make quantum mechanics feel necessary rather than merely true.

What follows takes that program and hangs it on a single hook — the most physical hook I can think of — and watches the whole structure follow. The hook is memory: the bare requirement that the universe be able to keep a fact around long enough to be checked.

The one thing we have to assume

Here is the entire foundation. We assume that the universe can hold a record: a fact that persists, that can be read back, that can be checked more than once and compared with other facts.

That sounds almost embarrassingly weak. But look at how much leans on it. A repeatable experiment is the assumption that the apparatus gives the same reading if you ask it again. A history of events is records of the past surviving to be compared with the present. Thermodynamics — the science of heat, engines, and the arrow of time — secretly depends on your being able to tell a fresh, blank memory from a used, full one. And an observer — anything that learns, including you — is by definition a thing that writes facts down and reads them back. Take away the assumption that records can persist and be re-read, and there is no experiment, no memory, no history, no thermodynamics, and nobody left to notice any of it.

So we grant it. We have no choice: any argument we could possibly make is itself a chain of stable, comparable facts, so to argue against records you’d have to use records. This is the one thing that cannot be derived from inside, because the deriving would already assume it. Everything else in this essay is the claim that this single grant, and almost nothing else, forces the rest of quantum mechanics into being.

A quick word on the honesty I promised. As we climb, I’ll tag each step:

• Granted — the one assumption above; we can’t prove it, we can only accept it.
• Theorem — a proper, proven result you could look up.
• Strong reconstruction — proven, but resting on one extra reasonable assumption I’ll name.
• Model-specific — shown to work in a particular detailed construction; suggestive, not yet general.
• Open — honest frontier; a good idea without a finished proof.

Now we climb.

Step one: a fact you can check twice must be one of a few clean alternatives

(Theorem.)

Start with the gentlest consequence. If a record can be read repeatedly and give the same answer each time, then its possible values must be perfectly distinguishable — sharply different, never blurring into one another. A reading that sometimes confused “yes” with “maybe” wouldn’t be a reliable record at all.

There’s a sharper edge here, and it has a famous name: the no-cloning theorem. You cannot make a perfect copy of an unknown quantum state unless its possibilities are perfectly distinguishable. (Intuitively: to copy a fact faithfully you must first read it faithfully, and you can only read faithfully what is sharply distinguishable.) Since a record worth the name has to be copyable — that’s what makes it a record others can share — its alternatives must be the clean, distinguishable kind. In the language physicists use, they must be orthogonal: as different from each other as it is possible to be, like the directions north, east, and up.

So already, from “a fact you can check twice and copy,” we have: a record is a choice among sharply distinct alternatives.

Step two: a repeatable question is a “filter that doesn’t smudge”

(Theorem.)

Now consider the act of asking — testing whether a record holds a particular value. A good test is one you can repeat: ask “is it value 3?” twice in a row and you’d better get the same answer both times, with the second asking changing nothing.

Mathematicians have a name for an operation that does nothing when you repeat it: it is idempotent (from Latin, “same power” — applying it twice equals applying it once). In quantum theory, a repeatable yes/no test that doesn’t smudge its own answer is described by an object called a projector. You can picture a projector as a perfect filter: it lets through exactly the part of reality that answers “yes” to one specific question, and asking again does nothing because everything that got through already says “yes.”

The point worth pausing on: we did not assume this projector business, the way a textbook would. We derived it from “the question gives the same answer if you ask it again.” The machinery is starting to build itself.

Step three: writing a fact down is reversible — collapse comes later

(Theorem.)

Here is where the standard story of measurement quietly goes wrong, and the record picture fixes it.

We are taught that measuring a quantum system “collapses” it — irreversibly, mysteriously, destroying information. But think about what writing a record actually is: you take a fresh, blank notebook and you copy a value into it, so that the notebook and the system now agree. That operation need not destroy anything at all. It is, in the precise mathematical sense, reversible — in principle you could run it backwards and un-write the note. (Physicists call the clean version of this an isometry, a map that embeds one space faithfully inside a larger one without losing anything; the relevant results are the Stinespring and Naimark “dilation” theorems, which say every measurement can be built this way.)

This is a genuinely liberating reframing. Recording is not collapse. When the universe writes down which alternative happened, it is just making a faithful, reversible copy, correlating the system with a register. The notorious “collapse” is a different event, and it happens later, for a different reason (Step eight below). At the moment of writing, nothing has been destroyed.

By the end of these first three steps we have, forced from the bare idea of a record: sharp alternatives, filter-like questions, and reversible copying into registers. This is the skeleton of quantum mechanics — without, yet, a single one of the famously weird features. Those come next, and they come from two further demands that any real record-keeper must meet.

Step four: why the universe runs on complex numbers

(Strong reconstruction.)

This is the first step that turns “record-keeping” into something recognizably quantum, so it deserves care.

Schoolroom numbers are the real numbers — the ordinary number line. But quantum mechanics insists on complex numbers, which extend the line with a second, “imaginary” axis (built on the square root of minus one). Students are told to accept this and move on. Why should reality need the larger number system?

The answer turns on a property with a forbidding name and a homely meaning: local tomography. “Tomography” just means reconstructing a whole from measurements of its parts (as a CT scan reconstructs a body from slices). Local tomography is the claim that you can know everything about a combined system — two particles, say — by having two observers measure their own particles separately and then compare notes. No spooky extra information that only appears when you look at both together; the parts, plus their correlations, tell the whole story.

That is exactly what record-keeping observers need. Each one interrogates their own corner of the world and writes down what they find; for them to reconstruct the joint reality, the joint reality had better be reconstructable from the separate pieces.

Now the magic. You can simply count how many numbers it takes to describe a system, and check which number system passes the local-tomography test:

• Complex quantum theory passes. The whole is exactly the parts and their correlations.
• Real quantum theory fails. It needs extra, global numbers that no separate, local measurement can ever reach — two different combined realities can look identical to every local observer. Records made by parties comparing notes would miss something real.
• Quaternionic theory (an even bigger number system) fails even to combine two systems consistently at all.

Add one more thing we already have — that records are written by smooth, reversible motion (Step three), which rules out a purely “classical” possibility — and a pair of theorems (Hardy’s and Chiribella’s) snap shut: the number system must be the complex one. The most notorious “we just have to use it” feature of quantum mechanics, the complex numbers, turns out to be the unique choice that lets local observers compare notes and miss nothing.

I’ve graded this a strong reconstruction rather than a flat theorem because it leans on one honest assumption: that the universe’s measurements really are local — built up from observations of small pieces, with no irreducibly all-at-once “global” measurement primitive. For a universe made of local cells that’s natural, but it’s an assumption, and I won’t pretend otherwise.

Step five: why a long-lived fact must be a protected fact

(Strong reconstruction.)

The second demand is the one our memory-keeping universe cannot dodge: the world is noisy. Finite, jiggling, bombarded by randomness. A record left to sit will degrade, like a sandcastle in the wind.

So a fact that must survive cannot just be written once and abandoned. It has to be actively defended. And the science of defending information against noise is error correction — the same idea that lets a scratched DVD still play and a deep-space probe still phone home. The trick, in its quantum form (here the relevant pioneers are Shor, Steane, Calderbank, and Gottesman), is wonderfully sly. You do not store your precious fact in any single place where noise could find it. You store it in the relationships among many places, and you check those relationships — physicists call the check a syndrome, as a doctor reads symptoms — without ever reading the fact itself. Noise trips an alarm (a violated relationship) while the protected information sits untouched.

Notice that this is precisely the non-disturbing, repeatable read our record demanded back in Step one. Error correction is not an optional engineering add-on bolted onto quantum mechanics; run the record requirements together and you are forced into the structure of an error-correcting code to keep a fact alive in noise.

And here the argument pays a startling dividend, worth a moment even in plain language. When you ask for the smallest code that balances the two kinds of error a record can suffer — corruption of what was written, and corruption of whether it was written — and that is stable under the basic operations a quantum world allows, the answer is forced to be a very specific little object: an eight-bit cell with the symmetry of a cube. There is no smaller one. (For the cognoscenti: it is the unique doubly-even self-dual code, the $[8,4,4]$, which sits exactly at the boundary between the “easy,” classically-simulable quantum operations and the genuinely powerful ones — the famous Clifford/non-Clifford line. The minimal record cell lives right on that edge.)

This is the deep heart of “it from bit.” We usually imagine the “bit” as a humble classical 0 or 1. Steps four and five say it cannot be. To compose, and to survive, the bit must be a protected quantum bit — a syndrome in an error-correcting code. The atom of reality, on this picture, is not a classical token. It is a self-defending quantum record.

Steps six through eight: where probability — and the famous “rule” — come from

(Theorems and model-specific results.)

With a complex, error-correcting record substrate in hand, the rest of quantum mechanics tumbles out, including the parts usually presented as separate miracles.

The stable states are the ones the code protects. A long-standing puzzle — “why does the world settle into these particular definite states and not others?” — dissolves. The states that survive the constant noise-checking are exactly the ones the error correction is built to preserve. Zurek called this einselection: the environment, by constantly monitoring, selects the survivors. In our picture the “environment” is just the error-correction process itself, perpetually reading the syndrome. The preferred states of the classical world are whatever the code defends.

The Born rule is the only consistent way to bet. This is the rule students are simply handed: the probability of an outcome is the square of an amplitude. Where does it come from? A 1957 theorem by Andrew Gleason answers it with surgical finality. Once you have filter-like questions whose odds don’t depend on which other (compatible) questions you happen to ask alongside them — a property called non-contextuality, which our error-correcting records automatically have — then there is exactly one mathematically consistent way to assign probabilities to them. One. No freedom. And it is the Born rule. The probabilities aren’t an extra postulate; they’re the unique consistent bookkeeping for record-questions. (Why the square, specifically? Because a probability here is the weight of a complete loop — write a record, then read that you wrote it — and the forward and backward halves multiply to give a squared magnitude.)

So: definite states, and the precise odds of finding each — both forced.

Step nine: the arrow of time is the cost of forgetting

(Model-specific, on a rock-solid foundation.)

Now we can finally locate the “collapse” that Step three refused. Reversible recording keeps filling registers with correlated copies. To go on recording — to make a reusable memory — you must eventually wipe registers clean. And erasing is where irreversibility lives.

This is not hand-waving; it is one of the most beautiful exact results in physics, Landauer’s principle (now confirmed in the laboratory): erasing one bit of memory costs a definite minimum amount of energy, released as heat. You can copy for free, in principle. You cannot forget for free. Ever.

So the thermodynamic arrow of time — the reason the future feels different from the past, the reason heat flows one way and eggs don’t unscramble — is, on this view, the running cost of clearing memory to make room for new records. And “measurement collapse” looks irreversible not because the writing was, but because the re-blanking is. The mystery of measurement turns out to be the mystery of the wastebasket.

Step ten — and the whole point: why none of this feels true

(Theorem, and the resolution of the “weirdness.”)

We have arrived at the step that explains everything that came before, including why you’ve spent your life thinking quantum mechanics is bizarre.

Why does the everyday world look classical — solid, definite, objective, the same for everyone — when its ingredients are these slippery quantum records? The answer is redundancy, and it was worked out under the name quantum Darwinism (Zurek again, with Ollivier and Poulin). A fact becomes an objective, classical, “just-there” fact when it gets copied redundantly into its surroundings — broadcast into the environment so many times that any number of observers can each grab a copy and all agree, without anyone disturbing the original.

Here is the thing to sit with. A macroscopic object cannot help but broadcast itself. Right now, the coffee cup on your desk is scattering trillions of photons every second, nudging air molecules, radiating warmth — and each of those photons flies off carrying a little copy of where the cup is. Its position is recorded, redundantly, in a vast cloud of escaping light and air, whether or not anyone is looking. You could not keep the cup’s location secret if you tried. The fact is over-copied, billions of times over, automatically.

And that is why the quantum rules are invisible to us. Our intuitions — our entire sense of what a “fact” is — were built, by evolution and a lifetime of experience, inside this over-copied regime. In our world, facts are cheap, public, and freely copyable. You can look at the cup without changing it (you’re just intercepting one of its zillion broadcast copies). Everyone sees the same cup (the copies all agree). So we grew up certain that this is simply how facts are: out there, readable at no cost, copyable without limit.

Quantum “weirdness” is nothing more than what reality looks like in the opposite regime — the regime where a fact exists in only one copy, not yet broadcast to anyone. For a single electron’s spin, a lone photon’s polarization, a fact the environment hasn’t yet copied a billionfold:

• you can’t copy it (no-cloning — Step one), because it isn’t the freely-distinguishable, already-broadcast kind;
• reading it disturbs it, because there’s no spare copy to intercept — you have to touch the only one there is;
• and until it’s been copied, it genuinely has no single definite value to report — that’s “superposition.”

None of that is spooky. It is exactly, precisely what the logic of records says must happen for a fact that has not yet been copied. The “weirdness” is just the unfamiliar regime our intuition never visited, because at our scale nothing stays uncopied for more than an instant. We are creatures of the redundant limit, mistaking our cozy corner for the whole of reality.

The classical world isn’t the normal one that quantum mechanics weirdly violates. The classical world is the special one — the over-copied limit — and quantum mechanics is the general rule it’s a special case of. We had it exactly backwards.

So: necessity, not contingency

Step back and see the shape of the climb. We granted one thing we cannot not grant — that the universe can keep a record. From there, step by step, much of it proven outright, we were pushed:

• into sharp, distinguishable alternatives (because records repeat);
• into filter-like questions (because tests repeat);
• into reversible writing (because copying needn’t destroy);
• into complex numbers (because observers compare notes);
• into error-correcting codes (because the world is noisy);
• into the Born rule (because there’s only one consistent way to bet);
• into a thermodynamic arrow (because forgetting costs);
• and into a classical, objective world (because facts get over-copied) — which is exactly the disguise that hides all of the above from us.

This is why I’d say quantum mechanics is closer to an epistemological necessity than a contingent discovery — a fact about what knowing and recording can possibly be, more than a fact about some particular stuff. A universe that contains observers — things that remember — has almost no freedom in how it must work. The rules we find so strange are the rules any rememberer is stuck with.

I want to keep my promise about honesty, because the honesty is what makes this more than a pretty story. Not every rung is equally secure. The first three steps and the Born-rule step are proper theorems. The complex-numbers and error-correction steps are strong reconstructions resting on one named assumption each (that measurement is local; that the code is of the minimal balanced kind). The classicality results are demonstrated in detailed models and are widely accepted in spirit. And the one thing under everything — that the universe keeps records at all — is granted, not proved, and cannot be proved without circularity. The genuinely open frontier isn’t any single rung; it’s the question upstream of all of them: why a record-keeping universe must take precisely these forms and not some other. That’s unfinished, and saying so is part of the point. A ladder is only trustworthy if you can see which rungs are bolted down.

A closing glimpse: the shape such a universe would have

(Frontier — the speculative end.)

If you take all this seriously and ask not just “how must such a universe behave” but “what would it be built out of,” something remarkable happens, and I’ll sketch it without pretending it’s settled.

Read the steps not as constraints but as building instructions. The minimal protected record cell — that little eight-bit, cube-symmetric object from Step five — becomes a literal brick. And when you ask what shape can carry it and tile space, the brick turns out to fit together only in three dimensions: the number of dimensions of space is not put in by hand, it falls out of the geometry of the smallest possible error-correcting memory. Stack the bricks, let information hop between them by the only rule that respects the symmetry (a specific, forced “shuffle” called a Grover coin), zoom out, and the hopping of records reproduces — this is the astonishing part — the equation Dirac wrote down for a relativistic electron. Particles, in this picture, are ripples in the universe’s record-keeping.

I flag all of that as frontier: it’s one specific construction, and turning “remarkable that it works” into “proven that it must” is exactly the unfinished work. The general ladder above doesn’t depend on it. But it hints that “it from bit” might be not only a statement about behavior but a recipe for a world — with space, time, and matter as the things a sufficiently careful record-keeper is forced to grow.

Coda

Wheeler liked to call the universe a “self-excited circuit” — a thing that brings itself into being by observing itself. Strip away the romance and a hard skeleton is left. Observation is record-keeping. Record-keeping in a noisy world demands error correction. Error correction demands complex amplitudes, definite outcomes, the Born rule, an arrow of time, and an objective classical surface — and that surface is the very thing that hides the rest from us.

It really may be from bit. Not as a slogan, and not by happy accident, but by a logic you cannot escape once you ask for a universe that remembers. The weirdness was never in the world. It was in our assumption that facts come free.

---

This is the popular-science companion to a more technical paper, “It from Bit, Rung by Rung,” which states each step as a graded claim with full references to the original results (Wheeler; Hardy; Chiribella–D’Ariano–Perinotti; Gleason; Zurek; Landauer; Stinespring; Gottesman) and works out the geometric construction in detail. It is archived on Zenodo.